From 4e470428d9456f42c6fe321e567f47a365dbb335 Mon Sep 17 00:00:00 2001
From: Quan HL <quan.luuhoang8@gmail.com>
Date: Thu, 12 Oct 2023 15:43:50 +0700
Subject: [PATCH] add eleven labs

---
 lib/routes/api/speech-credentials.js | 27 +++++++++++++++++++++++++--
 lib/utils/speech-utils.js            | 26 +++++++++++++++++++++++++-
 test/speech-credentials.js           | 23 +++++++++++++++++++++++
 3 files changed, 73 insertions(+), 3 deletions(-)

diff --git a/lib/routes/api/speech-credentials.js b/lib/routes/api/speech-credentials.js
index fd9d45e1..70f08c24 100644
--- a/lib/routes/api/speech-credentials.js
+++ b/lib/routes/api/speech-credentials.js
@@ -127,7 +127,8 @@ const encryptCredential = (obj) => {
     custom_stt_url,
     custom_tts_url,
     auth_token = '',
-    cobalt_server_uri
+    cobalt_server_uri,
+    model_id
   } = obj;
 
   switch (vendor) {
@@ -202,6 +203,12 @@ const encryptCredential = (obj) => {
       const cobaltData = JSON.stringify({cobalt_server_uri});
       return encrypt(cobaltData);
 
+    case 'elevenlabs':
+      assert(api_key, 'invalid elevenLabs speech credential: api_key is required');
+      assert(model_id, 'invalid elevenLabs speech credential: model_id is required');
+      const elevenlabsData = JSON.stringify({api_key, model_id});
+      return encrypt(elevenlabsData);
+
     default:
       if (vendor.startsWith('custom:')) {
         const customData = JSON.stringify({auth_token, custom_stt_url, custom_tts_url});
@@ -317,6 +324,10 @@ function decryptCredential(obj, credential, logger) {
   } else if ('soniox' === obj.vendor) {
     const o = JSON.parse(decrypt(credential));
     obj.api_key = obscureKey(o.api_key);
+  } else if ('elevenlabs' === obj.vendor) {
+    const o = JSON.parse(decrypt(credential));
+    obj.api_key = obscureKey(o.api_key);
+    obj.model_id = o.model_id;
   } else if (obj.vendor.startsWith('custom:')) {
     const o = JSON.parse(decrypt(credential));
     obj.auth_token = obscureKey(o.auth_token);
@@ -454,7 +465,7 @@ router.put('/:sid', async(req, res) => {
           custom_stt_endpoint_url,
           custom_stt_url,
           custom_tts_url,
-          cobalt_server_uri
+          cobalt_server_uri,
         } = req.body;
 
         const newCred = {
@@ -732,6 +743,18 @@ router.get('/:sid/test', async(req, res) => {
           SpeechCredential.sttTestResult(sid, false);
         }
       }
+    } else if (cred.vendor === 'elevenlabs') {
+      const {api_key, model_id} = credential;
+      if (cred.use_for_tts) {
+        try {
+          await testWellSaidTts(logger, {api_key, model_id});
+          results.tts.status = 'ok';
+          SpeechCredential.ttsTestResult(sid, true);
+        } catch (err) {
+          results.tts = {status: 'fail', reason: err.message};
+          SpeechCredential.ttsTestResult(sid, false);
+        }
+      }
     }
 
     res.status(200).json(results);
diff --git a/lib/utils/speech-utils.js b/lib/utils/speech-utils.js
index 0c6f8c28..31ca5ef7 100644
--- a/lib/utils/speech-utils.js
+++ b/lib/utils/speech-utils.js
@@ -203,6 +203,29 @@ const testWellSaidTts = async(logger, credentials) => {
   }
 };
 
+const testElevenlabs = async(logger, credentials) => {
+  const {api_key, model_id} = credentials;
+  try {
+    const post = bent('https://api.elevenlabs.io', 'POST', 'buffer', {
+      'xi-api-key': api_key,
+      'Accept': 'audio/mpeg',
+      'Content-Type': 'application/json'
+    });
+    const mp3 = await post('/v1/text-to-speech/21m00Tcm4TlvDq8ikWAM', {
+      text: 'Hello',
+      model_id,
+      voice_settings: {
+        stability: 0.5,
+        similarity_boost: 0.5
+      }
+    });
+    return mp3;
+  } catch (err) {
+    logger.info({err}, 'synthEvenlabs returned error');
+    throw err;
+  }
+};
+
 const testIbmTts = async(logger, getTtsVoices, credentials) => {
   const {tts_api_key, tts_region} = credentials;
   const voices = await getTtsVoices({vendor: 'ibm', credentials: {tts_api_key, tts_region}});
@@ -251,5 +274,6 @@ module.exports = {
   testDeepgramStt,
   testIbmTts,
   testIbmStt,
-  testSonioxStt
+  testSonioxStt,
+  testElevenlabs
 };
diff --git a/test/speech-credentials.js b/test/speech-credentials.js
index 19fe50b9..b2b54490 100644
--- a/test/speech-credentials.js
+++ b/test/speech-credentials.js
@@ -469,6 +469,29 @@ test('speech credentials tests', async(t) => {
     });
     t.ok(result.statusCode === 204, 'successfully deleted speech credential for Cobalt');
 
+    /* add a credential for elevenlabs */
+    result = await request.post(`/Accounts/${account_sid}/SpeechCredentials`, {
+      resolveWithFullResponse: true,
+      auth: authUser,
+      json: true,
+      body: {
+        vendor: 'elevenlabs',
+        use_for_stt: true,
+        use_for_tts: false,
+        api_key: 'asdasdasdasddsadasda',
+        model_id: 'eleven_multilingual_v2'
+      }
+    });
+    t.ok(result.statusCode === 201, 'successfully added speech credential for Cobalt');
+    const elevenlabs_sid = result.body.sid;
+
+    /* delete the credential */
+    result = await request.delete(`/Accounts/${account_sid}/SpeechCredentials/${elevenlabs_sid}`, {
+      auth: authUser,
+      resolveWithFullResponse: true,
+    });
+    t.ok(result.statusCode === 204, 'successfully deleted speech credential for Cobalt');
+
     await deleteObjectBySid(request, '/Accounts', account_sid);
     await deleteObjectBySid(request, '/ServiceProviders', service_provider_sid);
     t.end();